Character Recognition Systems A Guide for Students and Practitioners

by Cheriet, Mohamed; Kharma, Nawwaf; Liu, Cheng-Lin; Suen, Ching

• ISBN13: 9780471415701
• ISBN10: 0471415707
• Edition: 1st
• Format: Hardcover
• Copyright: 2007-10-12
• Publisher: Wiley-Interscience

Summary

Character Recognition Systems discusses character recognition, a process by which text-based input patterns produce meaningful output. Through the use of practical examples and theoretical concepts, the authors discuss feature extraction and selection, pattern recognition, as well as relevant issues such as the recognition of bank accounts, signature verification, and postal address recognition.

Author Biography

Mohamed Cheriet is Professor in the Department of Automation Engineering at the École de Technologie Supérieure of University of Quebec, Montreal. He is the Director of Synchromedia Consortium, working in the fields of image processing, document analysis and recognition, learning algorithms, perception, and intellipresence. Nawwaf Kharma is Associate Professor in the Department of Computer and Electrical Engineering at Concordia University, Montreal. He is Director of the Concordia Computational Intelligence Lab (CeCIL) and a member of the ACM-SIGEVO. ChenG-LIn Liu is a Research Professor and the Deputy Director of the National Laboratory of Pattern Recognition (NLPR), Institute of Automation, Chinese Academy of Sciences. He is working in the fields of pattern recognition, machine learning, and document analysis. Ching Y. Suen is the Director of the Centre for Pattern Recognition and Machine Intelligence of Concordia University in Montreal, working in the fields of handwriting recognition and human-computer communications.

Table of Contents

Preface	p. xiii
Acknowledgments	p. xvii
List of Figures	p. xix
List of Tables	p. xxvii
Acronyms	p. xxix
Introduction: Character Recognition, Evolution, and Development	p. 1
Generation and Recognition of Characters	p. 1
History of OCR	p. 2
Development of New Techniques	p. 3
Recent Trends and Movements	p. 3
Organization of the Remaining Chapters	p. 3
References	p. 4
Tools for Image Preprocessing	p. 5
Generic Form-Processing System	p. 5
A Stroke Model for Complex Background Elimination	p. 8
Global Gray Level Thresholding	p. 9
Local Gray Level Thresholding	p. 11
Local Feature Thresholding-Stroke-Based Model	p. 12
Choosing the Most Efficient Character Extraction Method	p. 15
Cleaning Up Form Items Using Stroke-Based Model	p. 19
A Scale-Space Approach for Visual Data Extraction	p. 21
Image Regularization	p. 22
Data Extraction	p. 24
Concluding Remarks	p. 29
Data Preprocessing	p. 30
Smoothing and Noise Removal	p. 30
Skew Detection and Correction	p. 32
Slant Correction	p. 34
Character Normalization	p. 36
Contour Tracing/Analysis	p. 41
Thinning	p. 45
Chapter Summary	p. 50
References	p. 51
Feature Extraction, Selection, and Creation	p. 54
Feature Extraction	p. 54
Moments	p. 55
Histogram	p. 58
Direction Features	p. 59
Image Registration	p. 64
Hough Transform	p. 68
Line-Based Representation	p. 70
Fourier Descriptors	p. 73
Shape Approximation	p. 76
Topological Features	p. 78
Linear Transforms	p. 79
Kernels	p. 86
Feature Selection for Pattern Classification	p. 90
Review of Feature Selection Methods	p. 90
Feature Creation for Pattern Classification	p. 104
Categories of Feature Creation	p. 104
Review of Feature Creation Methods	p. 105
Future Trends	p. 118
Chapter Summary	p. 120
References	p. 120
Pattern Classification Methods	p. 129
Overview of Classification Methods	p. 129
Statistical Methods	p. 131
Bayes Decision Theory	p. 131
Parametric Methods	p. 132
Nonparametric Methods	p. 138
Artificial Neural Networks	p. 142
Single-Layer Neural Network	p. 144
Multilayer Perceptron	p. 148
Radial Basis Function Network	p. 152
Polynomial Network	p. 155
Unsupervised Learning	p. 156
Learning Vector Quantization	p. 160
Support Vector Machines	p. 162
Maximal Margin Classifier	p. 163
Soft Margin and Kernels	p. 165
Implementation Issues	p. 166
Structural Pattern Recognition	p. 171
Attributed String Matching	p. 172
Attributed Graph Matching	p. 174
Combining Multiple Classifiers	p. 179
Problem Formulation	p. 180
Combining Discrete Outputs	p. 181
Combining Continuous Outputs	p. 183
Dynamic Classifier Selection	p. 190
Ensemble Generation	p. 190
A Concrete Example	p. 194
Chapter Summary	p. 197
References	p. 197
Word and String Recognition	p. 204
Introduction	p. 204
Character Segmentation	p. 206
Overview of Dissection Techniques	p. 207
Segmentation of Handwritten Digits	p. 210
Classification-Based String Recognition	p. 214
String Classification Model	p. 214
Classifier Design for String Recognition	p. 220
Search Strategies	p. 227
Strategies for Large Vocabulary	p. 234
HMM-Based Recognition	p. 237
Introduction to HMMs	p. 237
Theory and Implementation	p. 238
Application of HMMs to Text Recognition	p. 243
Implementation Issues	p. 244
Techniques for Improving HMMs' Performance	p. 247
Summary to HMM-Based Recognition	p. 250
Holistic Methods for Handwritten Word Recognition	p. 250
Introduction to Holistic Methods	p. 251
Overview of Holistic Methods	p. 255
Summary to Holistic Methods	p. 256
Chapter Summary	p. 256
References	p. 257
Case Studies	p. 263
Automatically Generating Pattern Recognizers with Evolutionary Computation	p. 263
Motivation	p. 264
Introduction	p. 264
Hunters and Prey	p. 266
Genetic Algorithm	p. 271
Experiments	p. 272
Analysis	p. 280
Future Directions	p. 281
Offline Handwritten Chinese Character Recognition	p. 282
Related Works	p. 283
System Overview	p. 285
Character Normalization	p. 286
Direction Feature Extraction	p. 289
Classification Methods	p. 293
Experiments	p. 293
Concluding Remarks	p. 301
Segmentation and Recognition of Handwritten Dates on Canadian Bank Cheques	p. 301
Introduction	p. 302
System Architecture	p. 303
Date Image Segmentation	p. 303
Date Image Recognition	p. 308
Experimental Results	p. 315
Concluding Remarks	p. 317
References	p. 317
Index	p. 321
Table of Contents provided by Ingram. All Rights Reserved.